DE69108723T2 - SUPERCONDUCTIVE WIRE. - Google Patents

Description

BACKGROUND OF THE INVENTION

To produce high quality, fine NbTi superconducting filament wire, manufacturers must rely heavily on the integrity of diffusion barriers. These barriers, typically made of Nb, are placed between the copper plating and the NbTi superconducting alloy that forms the bulk of a filament. The barrier serves to inhibit the formation of brittle CuTi intermetallics at the filament surface. Current processing schemes for producing NbTi superconducting wire, which contains 4,000 to 7,000 filaments, each approximately 6 µm in diameter at the final wire size, use Nb barriers that represent approximately 4% of the (non-copper) filament volume. The process typically involves exposing the barrier to temperatures of 500º to 800ºC for a period of about 15 hours, followed by three or more heat treatments of 40 to 80 hours at 300º to 450ºC as the wire is brought to its final size. In fine filament conductors made in this way, current densities (Jc) of over 2,750 A/mm2 at 5 Tesla (T) and 1,600 A/mm2 at 7 T (4.2ºK) can be achieved (see, for example, "Superconducting Wire and Cable for the Superconducting Super Collider", T.S. Kreilick, E. Gregory, D. Christopherson, G.P. Swenson and J. Wong, Supercollider 1, Plenum Press, 1989, 235-243).

Although not unacceptable, these current densities are well below the 3,800 A/mm² (5T, 4.2ºK) achieved in wires with larger diameter filaments. The reduction in current density with refinement of the Filaments are directly attributable to the problems with the diffusion barrier.

The long periods at high temperatures during processing contribute to the undermining of the Nb barrier due to Ti diffusion from the NbTi alloy core. Recall that as Ti diffuses into the pure Nb barrier, Ti depletion occurs at the surface of the NbTi alloy filament core, resulting in an overall lower Jc in the core. The diffusion of Ti through the barrier also leads to the formation of a CuTi compound at the surface of the filament. This brittle CuTi compound breaks during cold reduction, resulting in "nodules" that adversely affect both the Jc and the ductility of the wire. The interfaces between the copper matrix, the Nb barrier and the NbTi core in a filament play a central role in this process.

In cases where irregular interfaces occur, extensive local thinning of the diffusion barrier is observed. This effect is most pronounced in areas where there are projections into the barrier and even into the surrounding copper. It is believed that in these areas the CuTi compound is formed first and in a larger amount than in areas where such thinning does not occur. These CuTi areas, the nodules, break when the wire is pulled and subsequently cause pinching of the filament. As a result, the wire has a greatly reduced Jc performance and is prone to breakage.

The obvious solution to the problem of thinning the diffusion barriers is to simply use a thicker barrier so that there are no thin spots that cause problems. However, increasing the barrier thickness only comes at the expense of the NbTi. The barrier does not carry Jc at operating magnetic fields, so a larger thickness of the barrier reduces the overall Jc. In addition, increasing the barrier thickness does not solve the problem of Ti leaching from the NbTi. Although this is not as dramatic as the problem of CuTi compound formation, this problem cannot be ignored if the current capacity is to be improved. So, although increasing the barrier thickness is an obvious solution to the barrier thinning problem, it is far from ideal.

We have found that instead of compensating for surface irregularities by increasing the barrier, it is better to eliminate these irregularities, i.e. to form a surface behind the diffusion barrier that is as uniform as possible. If the diffusion barrier is surrounded by a uniform surface, it is more effective. The formation of nodules can be largely avoided, and if the barrier has to be made thicker, the necessary increase can be minimized. The production of uniform NbTi surfaces in filaments is the main object of the invention.

When two or more metals are processed simultaneously, the extent of the irregularities at their interface depends on many factors. These include:

1. The relative hardness of the materials;

2. The grain size within the materials;

3. The microstructure and overall structure within the materials as determined by previous metallurgical and mechanical processing (e.g. hot working, cold working and overall deformation).

From the point of view of uniformity, the ideal interface is an interface between two metals of approximately equal hardness, each having very fine grains aligned with the plane of the interface so that they have the smoothest possible surface. Obviously, rolled sheet materials meet this description very well. The use of rolled sheet for the purpose of increased filament uniformity forms the basis of the present invention. A previous patent issued to Supercon, Inc., No. 4,646,197, relates to the manufacture of Ta capacitor wire wherein fine grain Ta sheet was wound around a billet of Nb or Ta to provide a smooth wire surface with significant resistance to Ta grain growth along the wire at high temperatures.

DESCRIPTION OF THE INVENTION

For a better understanding of the invention, reference is made to the following description and examples, which refer to the accompanying drawings. In the drawing:

- Figure 1 shows the schematic shape of a monofilament block containing layers of a superconducting alloy according to the invention;

- Figure 2 shows the schematic design of a multifilament block using the monofilament produced according to the invention.

In the manufacture of fine superconducting filament wire, a monofilament wire is currently produced from which a multifilament wire is subsequently produced by performing one or more stacking operations. The monofilament wire is produced from an ingot which generally consists of three elements: a superconducting alloy ingot, a barrier material which is in the form of sheet metal wrapped around the ingot, and an extrusion sheath of copper which surrounds the barrier and the ingot. In the present invention, a fourth element is added: a series of of fine-grained layers (ASTM micrograin number > 8) of superconducting material placed between the barrier layers and the ingot (see Fig. 1). These superconducting layers reduce the extent of surface irregularities of the interface between the diffusion barrier and the ingot, and they are also believed to increase the Jc due to increased magnetic flux pinning provided by both the finer grain structure in the layers and the presence of the interfaces between the layers.

The superconducting alloy layer may form all or any portion of the superconducting cross-sectional area, the amount being determined by the level of uniformity required and the specific requirements of the application. The layer may be of any composition, with the restriction that it be made of the same base alloy as the central ingot. For example, if the ingot is Nb with 46.5 wt% Ti (NbTi46.5), the NbTi layer may be of the composition Nb with 50 wt% Ti (NbTi50). This, again, is determined by the application.

The smooth surface masks the fine irregularities of the ingot surface so that no local thinning of the diffusion barrier can occur as a result of grain penetration or inaccuracies left in the ingot manufacture. The ability to vary the composition of the layer also represents a great advantage in terms of diffusion losses. In the case of NbTi, the use of Ti-rich layers around the ingot serves to compensate for diffusion losses so that the Jc of the NbTi region is maintained. The use of higher Ti concentrations (approximately 50 wt% Ti) in the layer can also provide improved Jc performance. Ingots made of these 50 wt% Ti alloys are difficult to cold work, but this problem does not arise if the superconducting region is suitably between the Ti-rich layer and a less rich ingot.

Applying the present invention has shown another advantage in using the NbTi sheet. In summary, the experiment conducted consisted of making two monofilament blocks, one containing a standard ingot and the other containing NbTi sheets wound around a solid NbTi core. After processing, it was found that the wire containing the sheets had a higher current density than the solid NbTi material. This was independent of the size of the wire and throughout the magnetic field range of 1 to 9 T used in the Jc tests. The Jc in the sheet region was calculated based on the known block dimensions and was found to be 8-10% higher than solid NbTi. We therefore have reason to believe that the use of NbTi sheets can also increase the current density of a filament. This increase is probably due to an improvement in flux pinning caused by the interfaces between the layers; however, this is not yet known with certainty.

The present invention certainly represents a significant improvement in current barrier technology. Both the current demand for high quality superconducting NbTi wire with 6 um filaments and the magnet industry's demand for wire with filaments down to 2.5 um are satisfied by the application of the present invention. In particular, to produce the required 2.5 um material (23,000 filaments) additional stacking is necessary. This additional processing step applies even greater pressure to the barriers than is applied in the production of 6 um filament material. The present invention helps to maximize the effectiveness of the barriers so that with the The greatest effect can be achieved with the smallest amount of barrier material.

A preferred method of applying the invention is given in the following non-limiting example.

EXAMPLE I

Figure 1 shows a schematic cross-section of a NbTi monofilament ingot. The ingot is typical of NbTi monofilament, but NbTi50 layers have been included where a solid NbTi ingot would normally have been used. The Nb diffusion barrier and the NbTi layers each occupy approximately 4% of the non-copper volume. The NbTi layers have a composition of NbTi50 to compensate for Ti depletion during processing and to slightly increase the Jc in the finished conductor. The sheet is made from an ingot of NbTi50 by rolling and annealing it so that the grain size of the 0.38 mm (0.015 inch) thick sheet is below a nominal diameter of approximately 22 µm (ASTM micrograin size number > 8).

Three monofilament blocks must be made to provide enough material for the subsequent multifilament block. All of these blocks are made in the same way and processed in the same way. First, the NbTi ingots and the OFHC copper extrusion cladding, headers and tails are machined to the appropriate dimensions and the Nb and NbTi layers are cut to size. The standard monofilament design requires a 0.61 m (24 inch) core, so the NbTi ingots must be cut to this length and all the layer material to this width. Based on the design of Figure 1, each block requires one Nb layer with a length of 1.76 m (69.5 inches) and a thickness of 0.38 mm (0.015 inches) and one NbTi50 layer with a length of 1.73 m (68.0 inches) and a thickness of 0.38 mm (0.015 Inch). Splicing is easier if continuous strips are used, but some shorter sections can be used if necessary.

After all these parts have been cleaned, the monofilament blocks are assembled. Care must be taken to minimize voids in the layer area. The NbTi blank blocks are inserted last. The copper header and copper tailer are then attached to the top and bottom of each block by electron beam vacuum welding.

The encapsulated blocks are then prepared for extrusion by heating at 650º for 3 hours. The blocks are then extruded at 650ºC from a cladding diameter of 18 cm (7.125 inches) to a diameter of 2.54 cm (1.00 inches). The extrusions are then trimmed to remove excess copper at the head and tail ends and are then cold drawn to size for stacking: 2.6 mm (0.1033 inches) hexagonal wire from flat to flat.

Figure 2 shows schematically the multifilament block made from the hexagonal monofilament wire. Although the monofilaments are actually hexagonal, they are shown as circles in cross-section for simplicity. After a sufficient amount of OFHC copper is drawn and formed hexagonally, both the copper and NbTi monofilament wire are cut to lengths of 80.32 cm (31.625 inches) long. The sheath, head and tail ends are machined, all parts are cleaned, and then the block is assembled. The filaments are arranged symmetrically in the proportions shown in Figure 2, the intent being to minimize the void space that inevitably occurs when fitting a hexagonal, closely packed array into a round sheath. When the filament array is complete, the copper sheath is slid onto it. Additional copper filaments are added around the periphery of the array as necessary. added to ensure a firm pressing. The head and tail pieces are then attached to the casing by electron beam welding in a vacuum.

The multifilament block is then hot isostatically pressed to combine the various sub-elements. As a result of the monofilament design, the hot isostatically pressed multifilament block has a pitch to diameter ratio (s/d) of 0.168. Where diameter is the average NbTi filament diameter including its barrier and pitch is the distance between adjacent filaments. A low s/d ratio is important to prevent the filaments from pinching off during extrusion; when the filaments are tightly packed relative to their diameter, they provide mechanical support to each other. An s/d ratio of 0.168 is appropriate for a multifilament of this type.

After hot isostatic pressing, the head and tail of the multifilament billet are machined to achieve the sheath diameter after hot isostatic pressing. Next, the billet is heated and then extruded from a sheath diameter of 31.4 cm (12.36 inches) to a diameter of 8.25 cm (3.25 inches). The extruded product is cut and then drawn to the diameter where the heat treatment schedule begins. Using heat treatments separate from cold workings serves to optimize the Jc of the finished conductor. The heat treatment schedule can be changed for practical reasons, but the following schedule is typical for a fine filament NbTi conductor: 375ºC for 40 hours for each wire diameter of 2.54 cm (1.00 inch), 1.62 cm (0.640 inch), 1.17 cm (0.460 inch) and 0.82 cm (0.325 inch).

After the final heat treatment, the multifilament is drawn to the final diameter of 0.8 mm (0.0318 inches), with the filament diameter being approximately 6 µm. Such a conductor is well suited for use in advanced devices, such as the Superconducting Super Collider. The application of the invention should increase the Jc significantly above the 2,750 A/mm2 (5T, 4.2ºK) typically obtained with comparable fine NbTi filament conductors. The cut length (the average length of the wire segments that reach the final size) should also be significantly higher due to the higher ductility of the lump-free wire.

In accordance with the invention as described in Example 1, the titanium losses can be compensated by surrounding the monofilament NbTi ingot with NbTi layers having a higher titanium content than the ingot. However, this process requires the two alloys to be melted, which is expensive. Furthermore, alloys with a higher titanium content are more difficult to cold work than conventional alloys, so that the production of layers with a high titanium content is a problem.

An alternative to using layers of a high titanium alloy is to use layers that have the same composition as the ingot and alternate these layers with relatively thin layers of pure titanium. These thin titanium layers serve as a reservoir for titanium, which diffuses into the alloy layers as they lose titanium to the niobium barrier.

These titanium layers should generally be much thinner than the NbTi layers to promote complete diffusion of the titanium. A thickness ratio between NbTi layers and titanium layers of 7:1 or higher is preferred. The total volume of the NbTi and titanium layers should not exceed more than approximately 15% of the filament volume, as titanium depletion does not progress much further into the filament. Both the NbTi and titanium layers should have a nominal grain diameter of less than 22 µm (ASTM micrograin number > 8) to ensure maximum barrier effectiveness. The outermost titanium layer should lie within the outermost NbTi layer. The Nb barrier should therefore not be in direct contact with titanium, as this will only facilitate diffusion of titanium into the barrier.

Although the above discussion focuses on NbTi, it is obvious that these principles can be applied to any type 2 filament superconductor that experiences similar diffusion losses (e.g. NbZr, which loses Zr). This also applies to the following non-limiting example, which indicates a preferred method of carrying out the invention:

EXAMPLE II

Figure 3 shows a schematic cross-section of a NbTi monofilament ingot. The ingot is typical of NbTi monofilament, but with a series of layers of NbTi46.5 and pure titanium included where a solid NbTi ingot would normally have been used. The Nb diffusion barrier occupies approximately 4% of the non-copper volume, while the volume of the NbTi/titanium layer is approximately 5%. Both the NbTi and titanium ingots are manufactured from ingots by rolling and hot annealing in such a way that the nominal diameter of their grains is below approximately 22 µm (ASTM micrograin number > 8). The titanium ingot should have a purity of Grade 2 or higher for maximum machinability.

Three monofilament blocks must be produced to ensure there is enough material for the later multifilament block. All of these ingots are manufactured in the same way and processed in the same manner. First, the NbTi ingots and the OFHC copper extrusion claddings, headers and tails are machined to the appropriate dimensions and the niobium, NbTi and titanium layers are cut to size. The standard monofilament design requires a 0.61 m (24 inch) long core, so the NbTi ingots must be cut to this length and all the layer material to this width. Based on the design of Figure 3, each block requires one Nb layer 1.83 m (72.0 inches) long and 0.38 mm (0.015 inches) thick, one NbTi46.5 layer 1.73 m (68.0 inches) long and 0.38 mm (0.015 inches) thick, and one Titanium layer 1.73 m (68 inches) long and 0.05 mm (0.002 inches) thick. Assembly is easier if continuous strips are used, but several shorter sections can be used if necessary.

After all of these parts have been cleaned, the monofilament blocks are assembled. Care must be taken to minimize voids in the area of the layers. The NbTi blank blocks are inserted last. The copper header and copper tailer are then attached to the top and bottom of each block by electron beam welding in vacuum. The encapsulated blocks are then prepared for extrusion by heating at 650º for 3 hours. The blocks are then extruded at 650ºC from a cladding diameter of 18 cm (7.125 inches) to a diameter of 2.54 cm (1.00 inches). The extrusions are then trimmed to remove excess copper at the head and tails and they are then cold drawn to size for stacking: hexagonal wire 2.6 mm (0.1033 inches) flat to flat.

The product is then treated as shown above in connection with Figure 2. After a sufficient amount of OFHC copper has been drawn and hexagonally formed, Both the copper and NbTi monofilament wire are cut to 80.32 cm (31.625 inch) lengths. The sheath, head and tail pieces are machined, all parts are cleaned, and then the block is assembled. The filaments are arranged symmetrically in the proportions shown in Figure 2, the intention being to minimize the void space that inevitably occurs when a hexagonal, closely packed array is fitted into a round sheath. When the filament array is complete, the copper sheath is slid onto it. Additional copper filaments are added around the periphery of the array as necessary to ensure a tight seal. The head and tail pieces are then attached to the sheath by vacuum electron beam welding.

The multifilament block is then hot isostatically pressed to combine the various sub-elements. As a result of the monofilament design, the hot isostatically pressed multifilament block has a pitch to diameter ratio (s/d) of 0.168. Where diameter is the average NbTi filament diameter including its barrier and pitch is the distance between adjacent filaments. A low s/d ratio is important to prevent the filaments from pinching off during extrusion; when the filaments are tightly packed relative to their diameter, they provide mechanical support to each other. An s/d ratio of 0.168 is appropriate for a multifilament of this type.

After hot isostatic pressing, the head and tail of the multifilament billet are machined to achieve the sheath diameter after hot isostatic pressing. Next, the billet is heated and then extruded from a sheath diameter of 31.4 cm (12.36 inches) to a diameter of 8.25 cm (3.25 inches). The extruded product is cut off and then drawn to the diameter at which the heat treatment schedule begins. Using heat treatments separate from cold working is to optimize the Jc of the finished conductor. The heat treatment schedule may be varied for practical reasons, but the following schedule is typical for a fine filament NbTi conductor: 375ºC for 40 hours for wire diameters of 2.54 cm (1.00 inch), 1.62 cm (0.640 inch), 1.17 cm (0.460 inch), and 0.82 cm (0.325 inch).

After the final heat treatment, the multifilament wire is drawn to the final diameter of 0.8 mm (0.0318 inch), with the filament diameter being approximately 6 µm. Such a conductor is well suited for use in advanced devices such as the superconducting super collider. Application of the embodiment of Example II should increase the Jc significantly above the 2,750 A/mm2 (ST, 4.2ºK) typically obtained with comparable fine NbTi filament conductors. The cut length (the average length of the wire segments reaching the final size) should also be significantly higher due to the higher ductility of the lump-free wire.

Claims (9)

1. Superconducting filament comprising a core of a superconducting NbTi alloy surrounded by at least two separate layers of a NbTi alloy, a barrier of refractory metal surrounding the layers and a shell of normal metal surrounding the barrier, the shell, the layers and the core being metallurgically bonded together. 2. Superconducting filament according to claim 1, wherein the interfaces between the layers of NbTi alloy are spaced apart by less than 0.5 µm. 3. A superconducting filament according to claim 1, wherein the Ti concentration in at least one of the separated layers is higher than in the core. 4. Method for producing a superconducting filament with the following steps: a core made of a superconducting NbTi alloy is surrounded by at least two separate layers made of an NbTi alloy, a high-melting metal barrier surrounding the layers and a shell made of normal metal surrounding the barrier are attached, and the shell, the layers and the core are metallurgically bonded to one another by a common reduction of at least 10 times. 5. The method of claim 4, further comprising the step of carrying out a plurality of heat treatments of over 450°C with a total duration of at least 10 hours. 6. The method of claim 4, wherein the separated layers of NbTi alloy have an initial grain size of less than 22 µm nominal diameter. 7. Method according to claim 6, in which the layers of the fine-grained NbTi layer are interspersed with at least one layer of one of the metals Nb or Ti. 8. A method of making a monofilament of a superconducting NbTi alloy comprising the steps of: surrounding an ingot of a superconducting NbTi alloy with at least two layers of a fine-grained superconducting NbTi layer and an outer barrier layer of a refractory metal, surrounding the barrier with a copper sheath, the fine-grained layers having a grain size of less than 22 µm in diameter, hot-extruding the resulting product, and reducing the extruded product to wire having a filament diameter of less than 10 µm, so that each of the NbTi layers is less than 0.5 µm thick. 9. Method according to claim 8, in which the layers of the fine-grained NbTi layer are interspersed with at least one Ti layer.